Fig1. a) Cross section TEM of a FePt/MgO(3 nm)/FePt MTJ. b) Hysteresys loop recorded on a similar structure that demonstrates decoupling of the two electrodes.
Using our knowledge of high perpendicular magnetic anisotropy system we have grown epitaxial FePt/MgO/FePt tunnel junctions. In this system, the coupling induced by the stray field from propagating domain wall is detrimental in coupling the two electrodes as demonstrated by repeated cycling of the soft layer. In another system, we have demonstrated spin injection into silicon from fully remanent Pt/Co electrode.
Spintronic was born 20 years ago with the transistor of Datta and Das. After many setbacks, it was recognized that spin injection from a ferromagnetic electrode to a semiconductor is only possible by tunnelling, either trough a tunnel barrier or a Schottky barrier. In the meantime people also addressed the way to correctly detect successful injection. Two paths were investigated: electrical detection (like in GMR) or optical detection (by using a p-i-n diode called a “spin-led” were polarized spins recombine with holes to emit circularly polarized photons). The latter was the most fruitful and many works have been published on that issue. However in order to analyze correctly and quantitatively the injection results, the axis of quantization of the spin has to be perpendicular to the plan of the quantum wells of the p-i-n diode where the recombination takes place. This is done by applying a magnetic field to saturate the ferromagnetic electrode. We chose to use high perpendicular anisotropy electrode L10 alloy like FePt, for which our laboratory has a tremendous knowledge. We reckoned that injecting from a fully remanent electrode would be a unique advantage for studying fundamental properties. Concerning the tunnelling barrier we chose epitaxial MgO because of its unique spin-filtering properties.
We chose GaAs because its direct gap enables a straightforward relation between circular dichroism and spin injection polarization. We have developed the epitaxial growth of MgO tunnel barrier on GaAs. In order to measure the spin polarization of the FePt/MgO system we studied FePt/MgO/FePt magnetic tunnel junction. Fig.1 shows that we managed to grow fully epitaxial tunnel junction with L10 electrodes. The bottom electrode grows fully strained on the buffer whereas the top electrode is completely relaxed through interface dislocations at the interface with the MgO tunnel barrier. This system shows on a macroscopic scale anti-parallel state suitable for MTJ .
Fig. 3 a) SiGe quantum well (QW) electroluminescence measured at 5K and 77K and at remanence. b) Circular polarization from the QW emission shows 100 % remanence demonstrated that the recombining electrons come from the ferromagnetic electrode.
Few studies have been reported on spin injection in Si. Main reason is the optical detection because of the indirect gap and low spin-orbit coupling. We chose a Si-Ge quantum well embedded in a Si p-i-n diode for the detection to remedy those defects. The injector is a Pt/Co/AlOx electrode, annealed at 450 °C to ensure perpendicular magnetization of the Co layer. Fig. 2 shows that we succeeded in injecting spin polarized electrons into Si from a fully remanent electrode. Circular polarization as a function of the field matches the hysteresis loop of the electrode. Dichroism remains constant up to 200 K.
 de Person et al., Phys. Rev. B, 76 184402 (2007).
 L. Grenet et al., Appl. Phys. Lett., 94 032502 (2008).
Maj : 17/10/2013 (479)